LESSONS FROM COUNTRIES

Transcription

1 PART III LESSONS FROM COUNTRIES WITH NON-FINANCIAL DEFINED CONTRIBUTION SCHEMES 395

2

3 Chapter 15 NDC Strategy in Latvia: Implementation and Prospects for the Future Edward Palmer, Sandra Stabin, a, Ingemar Svensson, and Inta Vanovska* INSPIRED BY THE REFORM LEGISLATION PASSED IN SWEDEN IN 1994, Latvia formulated and passed its own NDC legislation in In doing so, Latvia became the first country in Central and Eastern Europe to legislate an NDC reform. With the implementation of its reform in January 1996, it was the first country to make a complete transition to NDC for the entire working population. This study tells the story of how the Latvian pension reform evolved, discusses problems encountered along the road of its implementation, assesses its results, and discusses the main lessons to be learned from the Latvian experience. The reform set a contribution rate of 20 percent for non-financial defined contribution 1 (notional) accounts from 1996 and for notional accounts and financial accounts together from July 2001, with a planned gradual transition to a split of the 20 percent contribution rate between non-financial (NDC) and financial defined contribution (FDC) pension schemes by The cost of old age pensions at the time of the reform was well over 20 percent; since a contribution rate of only 20 percent gives total pension rights to current workers, contributions above this level are viewed as a tax. The tax covers the transition from the old to the new regime and also finances a minimum guarantee. All of the main bolts of the Latvian Pension reform were in place in 2001 with the introduction of the individual financial account scheme. The reform promises stability in the face of demographic and economic fluctuations and manages a broad range of possible economic and demographic scenarios or risks that the country may have to face. It provides an adequate average pension level for career workers and guarantees an acceptable minimum pension standard. The reform path was bumpy, however. The conversion to NDC in 1996 brought a number of transition issues that had to be resolved; some of these issues had to be revisited in the period immediately following the initial reform legislation. The most important conversion issues involved the questions of how to value rights acquired under the old regime and how to introduce the system in an economic environment characterized by structural upheaval as the country was just beginning the process of transition from a command to a market economy. * Edward Palmer is at the Swedish Social Insurance Agency and the University of Uppsala; Ingemar Svensson is at the Swedish Social Insurance Agency; and Sandra Stabin, a and Inta Vanovska are at the Ministry of Welfare of Latvia. The model used for calculations in this study is the official model used by the Latvian government for the analysis of medium- and long-term pension policy. 397

4 398 PENSION REFORM: ISSUES AND PROSPECTS FOR NDC SCHEMES Setting a reasonable minimum pension age for the new system was a major systemic issue in Latvia, but also a political one. With the actuarial character of NDC and FDC it became important to set a high enough pension age so that persons leaving the labor force at the minimum age would receive an adequate pension. However, in accordance with the rules in the old regime, workers especially women were used to receiving a benefit at a young age: 55 years old for women and 60 for men. For this reason, raising the pension age was a sensitive political issue. A third important issue was how to phase out widespread special privileges dating from the pretransition period for listed groups of persons below the prereform minimum pension ages of 55 for women and 60 for men. The final major issue, which was also a topic of discussion during many years after 1996, was that of phasing in the mandatory financial account scheme, given the alternative goals of reducing the contribution rate and phasing in wage indexation of pension benefits. How these issues were dealt with and the loose ends that still remain at the time of this writing are among the topics of this study. We begin with a brief overview of the economic environment prevailing as the government began to consider the reform and review the reasons for why reform was necessary. The following section summarizes the reform legislation. Then we devote separate sections to discussing the transition issues and transition process during the initial years, and to discussing the design of the overall reform, including the phasing in of the FDC scheme, and the issue of long-term financial stability. The final section addresses the questions of political sustainability and popular acceptance. The Early Transition Years and Picture Prior to the 1996 Reform After gaining independence in 1990, Latvia moved with determination toward establishing a market economy and democratic institutions. As in other transition countries, this process led to a collapse in the economy. At its worst, in 1992, prices skyrocketed by 900 percent. Latvia responded with resolve to the transition challenge, and by 1993 the currency was stabilized and structural reforms were under way, trade was open, and prices were largely deregulated. Nevertheless, in the initial years of the transition, , GDP fell by more than 40 percent. As the bottom fell out of the economy, enterprises produced articles no longer marketable. One of the harsh but necessary tasks of the transition was to dissolve unproductive enterprises and privatize enterprises that had some market value in the new economy. This process of necessary structural change led to disrupted working careers for a large percentage of the Latvian population. By 1995, only a few large enterprises remained to be fully privatized. Latvia had become a nascent market economy, but, as it looked in 1995, at a high price. In 1995, as the pension reform was being formulated, there was still no sign that the economy would rebound into sustained growth, unemployment was high, and many of those who did have work did not have fulltime employment. As pension expenditures were largely fixed, the 40 percent fall in GDP and the wage base meant that the ratio of pension expenditures to GDP went from 5.5 in 1985 to around 10.5 percent in the mid-1990s. 2 This was precisely when a relative increase in budget expenditures was most difficult to afford. Expenditures of a little more than 10 percent of GDP on old age pensions may not seem alarming gauged by Western European standards; however, at the time, given all the other demands on dwindling public revenues and the difficulties people had in just making ends meet for the day, this was high. Per capita income had fallen dramatically and poverty was widespread. In addition, pensioners were generally no worse off than families with children, and persons in the rural areas were struggling because of the collapse of the farming infrastructure. 3

5 NDC STRATEGY IN LATVIA: IMPLEMENTATION AND PROSPECTS FOR THE FUTURE 399 From 1991 to 1995, the number of persons for whom contributions were being paid declined by almost 50 percent, and, by the end of 1995, there were about 1.5 persons receiving a pension per contributor. 4 So, the point of departure for the reform was that there were too many people receiving benefits and too few paying contributions. The first problem the large number of pensioners was created by the especially low pension age of 55 for women and an extensive list of occupations and circumstances that entitled people to benefits as early as age 40. The rules were inherited from the old Soviet system, in which there was a proclivity to use pensions promises of remuneration in the future not only to compensate for harsh or dangerous working conditions (rather than to improve them) but also to provide special privileges for persons in, for example, the police, military, and judiciary occupations; to reward mothers giving birth to a large number of children; and so on. By the 1990s, there was a long list of occupations and circumstances for which there were special pension privileges, since this was a way to negotiate relative income advantages under the old regime. The second problem too few contributors was, first, in part simply the other side of the early retirement coin. But it was also exacerbated by unfavorable demography. In addition, it was a manifestation of the informal economy that had blossomed in the transition. Deals were made between employers and employees where minimum wages were declared for revenue collection purposes, and remuneration above this was paid under the table. For those who worked where the opportunity to avoid paying contributions was present and wages were low, evasion or making a payment based on a minimum wage appeared to be a rational form of behavior. Those penalized by this state of affairs were persons within the system, those who had little or no opportunity to evade paying social insurance contributions for example, employees in large companies and civil servants. Most importantly there was no link between contributions and benefits in the old pension formula, and no actuarial adjustment for postponed retirement. For all of these reasons, the Latvian pension system was in crisis when the reform concept was formulated in In sum, the rules of the old defined benefit system were not appropriate for a market setting and an aging population. The old system had created and would continue to create too many pensioners and too few contributors. Moreover, there were weak incentives to pay contributions and no incentives to postpone collecting pension benefits past the minimum pension age. Foremost on the agenda of policy makers was an increase in the minimum pension age, and thereafter a change in the benefit formulas leading to a combination of universal, mandatory non-financial and financial pension schemes based on lifetime accounts and life expectancy, with a direct link between contributions and benefits. The Latvian Reform The Latvian reform of the statutory old-age pension system, formulated and legislated in 1995, consists of two components. A non-financial defined contribution (NDC) component was implemented on January 1, 1996; since July 1, 2001, a financial defined contribution (FDC) component has also been part of the new system. In the first component, individuals have notional individual accounts, whereas in the second they have individual financial accounts. There is also a general and a temporary transition guarantee in the form of a minimum pension for persons who reach the minimum pension age. This is financed by revenues outside the overall contribution rate of 20 percent for the NDC and FDC schemes, together. As has already been mentioned, during the transition, the total cost of paying benefits granted under the old system and covering the introduction of the FDC scheme is over 20 percent, with the excess being regarded as a tax to finance the transition.

6 400 PENSION REFORM: ISSUES AND PROSPECTS FOR NDC SCHEMES The transition from defined benefit scheme(s) to defined contribution schemes introduced the principle that the present value of an individual s benefit is equal to the present value of his or her contributions, and the systemic restriction that the present value of total assets in the system must be at least equal the present value of total liabilities. The latter condition is fulfilled automatically in an FDC scheme, and is fulfilled in the NDC scheme given that the rate of return on credited accounts is not greater than that necessary to fulfill this condition. 5 Here we describe in greater detail how the two schemes are set up. The NDC Component of the Reform The NDC scheme covers all persons living and working in Latvia from age 15, including the self-employed and farmers. Rights acquired in the old scheme were converted to NDC capital beginning January 1, 1996, and persons retiring in 1996 and after have retired according to the new rules, with NDC capital computed with a transition formula (described in detail below). The details of the conversion to NDC are discussed below. The contribution rate on earnings noted on individual accounts was set at 20 percent from January 1, Accumulated notional pension capital is determined by contributions, paid on all earnings up to a ceiling (in 2004 about 10 times the average wage), and during a whole working life. As long as one works, one pays contributions and these are noted on individual accounts. A benefit can be claimed at any time from the minimum pension age (see below), and it is possible to combine work with an NDC annuity after reaching the minimum pension age. The principle behind this is that it provides an opportunity and support for gradual withdrawal from the labor force. Noncontributory rights in conjunction with childcare and military service are financed with transfers from the state general budget. Rights in the old-age pension system derived during periods covered by social insurance for unemployment, compensated sickness absence of employees and the self-employed, work injury, and disability are financed by transfers from the other accounts in the overall social insurance budget to pension accounts. 6 The rate of return on notional capital is set equal to the growth of contributions, that is, the contribution wage sum. This arrangement has two features worth pointing out. The first is that the rate of return earned during the accumulation period will reflect not only the growth of productivity and the wage rate, but also the effect of changes in fertility, the net migration of working age persons, and the gradual transition from the informal to the formal economy. Productivity and wage growth can be expected to be high in Latvia in the coming two decades, whereas fertility is expected to remain low. What s more, well-educated and well-trained younger persons are expected to leave Latvia to seek better labor market prospects abroad. Finally, with time, market formality will increase, providing a broader contribution base. Fertility dropped to around 1.2 children per woman in the mid-1990s and has remained more or less at this level up to the time of this writing, in Demographers do not believe that the fertility rate will reach the level of 2.1 or higher needed to reproduce the population in the coming two decades even in the most optimistic of scenarios. 7 For the birth rate to once again become so high, it is important for younger Latvians to feel that they have achieved a sufficient level of economic and job market security, and that they can look forward to a living standard sufficient to support larger families. Furthermore, transfers to parents will compete for some time to come with transfers to children within the family. In the 1990s, the shadow economy was estimated to account for 30 to 35 percent of total national production. There are two sources of underreporting of income. One is that some

7 NDC STRATEGY IN LATVIA: IMPLEMENTATION AND PROSPECTS FOR THE FUTURE 401 workers just don t report earnings at all for some hours worked to the tax authority, which is a labor component and has an effect similar to having a smaller labor force. The other is the practice of not reporting a full salary, but, for example, reporting a salary close to the minimum wage. This is a per capita wage component that has the same effect as a lower wage per capita. Both components of informality are expected to improve over the coming decades as the economy evolves into an overt market economy. As this happens, these gains will also be reflected in the overall scale of the system, and the long-term ability of the system to finance better benefits. In sum, the gradual transition from the informal to the formal market economy will contribute to formal growth in contributions and constitute a separate component of the rate of return on current NDC accounts of workers and pensions, the latter through the partial real indexation formula of the Latvian system (see below). To date the real average annual rate of growth of the covered wage sum has been 6.7 percent, with annual growth of (covered) per capita real wages of 5.5 percent on average from 1997 through The real rate of return on individual notional accounts was, thus, 6.7 percent during The number of contributors grew by 1.1 percent during the same period. The rate of inflation was about 3.6 percent. The nominal rate of return on capital was, hence, 10.5 percent. The key parameters for NDC pensions for the period are summarized in table The growth in the number of contributors at an average annual rate of 1.1 percent is particularly noteworthy. This is consistent with the idea that the introduction of NDC (and FDC) is at least neutral in terms of promoting formality, and also supports the idea that it can have a positive effect. However, it is difficult to separate this effect from an overall effect of successful transition from dishevel to a formal market setting. These processes go hand in hand. At the least, the introduction of NDC in 1996 provided a logical framework for a system moving in the direction of formality. What turned out to be important for the rate of return in the NDC scheme in the first decade of its existence was the high rate of growth of real wages per capita at an annual average of 5.5 percent in spite of the negative repercussion of the Russian Crisis of Overall, growth in contributors and real wages provided an overall real rate of return of 6.7 percent for NDC accounts. As table 15.2 indicates, the percentage of economically active persons contributing increased dramatically from 1997 through Again, this was probably a general effect of the realization of the transition, but it was also encouraged by a social insurance system where benefits are tightly linked to individual contributions, a major message propagated to the public since the pension reform came in The positive effect generated by the growth of contributors will continue as long as the positive impact of increasing formality surpasses the negative impact of a declining population according to present demographic projections the labor force will decrease at a rate of a half percent per year or more (see below). The scales are likely to be tipped Table Overview of Some Key Economic Indicators for NDC Pensions 1. Real per capita wage growth, average % 2. Growth of contributors, average % 3. Growth of real wage sum, average % 4. Rate of inflation, average % 5. Career years for newly granted benefits in 2003 (1996) 31 (30) 6. Unisex life expectancy at age 60 in 2002 (1996) (17.88) Source: Ministry of Welfare of Latvia (2003).

8 402 PENSION REFORM: ISSUES AND PROSPECTS FOR NDC SCHEMES Table Key Demographic Pension Ratios Ratio (1) Ratio of working-age population (20 59) to population age (2) Ratio of persons in employment to population age 60 + excluding job seekers (3) Ratio of contributors to economically active persons (4) Ratio of contributors to population age (5) Ratio of contributors to old age pensioners (6) Ratio of contributors to old age and disability pensioners Source: Ministry of Welfare of Latvia. Prepared for this study. toward a net negative process, which can be expected at latest around 2020, but quite possibly earlier given the rapid pace in the increase in formality since The rate of growth of real per capita wages is also likely to subside well before the 2030s. Hence workers around and older around the year 2000 will be the recipients of this growth when they become pensioners. Younger birth cohorts, on the other hand, are covered increasingly by the FDC scheme and will rely much more on this scheme rather than on NDC for their future benefits. Also, as calculations presented in a separate section in this study indicate, the introduction of an NDC reserve fund is essential to maintain financial balance in the coming 40 years. Latvia does not have the Swedish automatic balancing mechanism. If it were to be introduced in Latvia it would most likely reduce NDC commitments even further than in the calculations presented in this study for the most pessimistic scenarios those where fertility, net immigration, mortality, and real per capita wage growth all move against the system. On the other hand, such a balancing mechanism would not be needed in practice to maintain a ratio of assets to liabilities greater than unity financial balance in scenarios with non-pessimistic economic and demographic outcomes. Generally speaking, the NDC scheme can also afford indexation of benefits with the rate of growth of the contribution base. This was not possible to implement in practice, however. In the beginning of the reform in 1996, it was only possible to price-index (all) benefits. Then the stock of pensioners consisted almost solely of persons whose benefits had been determined under the old rules, including the special rights giving benefits from age 40. Largely because of this generosity in granting pension benefits, but also because of the fall of GDP during the early 1990s, pension costs at the time of the reform were already too high. Pensioners will increasingly become NDC pensioners and eventually the system can move toward including indexation based on real growth. It is also important to note that the historical demographics of Latvia also work in favor of NDC system implementation, since large cohorts born in the 1980s will be entering the labor force in the period However, this also means that internal financial equilibrium in the system requires demographic reserves. In sum, it will be possible to move gradually in the direction of full wage indexation, but a course should be set out that also takes the need for reserves into consideration. The Ministry of Welfare has recently performed calculations 8 similar to those presented below in this study that show how this path could be taken. In fact, in principle, the Latvian NDC scheme is overfinanced given the counterfactual of possible full-fledged NDC indexation for NDC pensioners. The introduction of the FDC scheme in 2001 was given priority over full indexation of NDC benefits, as both were not affordable in parallel during the initial years. Until 2002 the indexation of benefits was based solely on the consumer price index. In , pen-

9 NDC STRATEGY IN LATVIA: IMPLEMENTATION AND PROSPECTS FOR THE FUTURE 403 sions with small amounts were indexed on the basis of both the change in the Consumer Price Index and 25 percent of the real growth of contribution wage sum. From 2004 until 2006, small pensions (that is, pensions not exceeding three social benefits 9 or 105 lats in 2004) will receive 50 percent real indexation in addition to inflation adjustment. According to the current pension law in , all pensions with an amount not exceeding five social benefits will receive 25 percent real indexation and after that the real growth component will be increased to 50 percent for all pensioners. The annual pension (P) at retirement in the Latvian NDC pension scheme is calculated as P = K/G, where K is the accumulated life-time notional pension capital at retirement and G is cohort unisex life expectancy at retirement. G values are calculated from projected cohort life tables. 10 Assumptions for the projections are prepared and analyzed by an official committee of Latvian demographic experts, including the experts from the Statistical Bureau. A report from mortality experts is discussed and approved in the committee and delivered to the Ministry of Welfare. The initial set of cohort projections were determined by examining mortality trends in countries with much lower mortality rates than Latvia and then judging probable paths of convergence of Latvian mortality to better practice. The value of G is set annually, by the Ministry of Welfare for retirement ages between 40 and 80, 11 in accordance with the recommendation of the committee. In 2004, the G value for retirement at ages 60, 65, and 70 were 19.7, 15.67, and 12.46, respectively. According to the present projection, the G value at age 65 is expected to increase by about 2 years during the coming 30 years. The long-term projections of mortality should change only gradually as new evidence becomes available. More importantly, even large changes in current period mortality rates will have no effect on the G values of future retirees, as long as the projected long-term mortality trend is not revised. There were two arguments in favor of using cohort projections for Latvia. The first is a general, not country-specific argument: Compared with using recent historical cross-sectional data, cohort trend-based projections are expected to come close to the actual development of future mortality, and should minimize the gap between the ex ante mortality rate used to compute the annuity at retirement and the ex post outcome, viewed when participants have deceased some time in the future. The second argument in favor of cohort projections is relevant for many transition countries. If the G values had been based on a procedure involving recent current history, this would have led to undesirable fluctuations in G values from one year to the next. As figure 15.1 shows, there have been very large and rapid changes in mortality rates in Latvia since the mid-1980s. Between 1985 and 1987, Gorbachev s antialcohol campaign led to a sharp rise in life expectancy for men, but this was followed by a relapse in the late 1980s. The situation then deteriorated even more during the break-up of the Soviet Union and the transition period to the market economy in the 1990s. Between 1987 and 1994, unisex life expectancy at birth fell by five years. In the high ages, which are directly relevant for G-value calculation, there was an extreme increase in mortality during the early and mid-1990s, probably as a consequence of the economic hardship and disruption in health care created by the transition. These changes were not seen as relevant for younger cohorts claiming an annuity after On top of this, in 1996 life expectancy at birth increased by 3.2 years for males and 2.5 years for females (see figure 15.1) based on period life tables that is, data representing current mortality rates. It would not have been reasonable to construct longevity projections on the basis of these short-term historic down and upturns. Hence, for this reason too cohort projections were preferable. In sum, the method used to determine the rate of return on NDC capital balances during the accumulation phase, along with the forward-looking cohort mortality estimate

10 404 PENSION REFORM: ISSUES AND PROSPECTS FOR NDC SCHEMES Figure Life Expectancy at Birth in Latvia E(0), years females both sexes males Source: Krumins (2000). used in the calculation of the annuity, contribute to creating long-term financial stability. The data needed to create an NDC financial balance along the lines created in Sweden are still not available. Among other things, in the initial years of the reform initial capital was computed and registered only for persons claiming benefits. The process of registering transition notional capital for all, which involves individuals turning in their workbooks from the Soviet times to the Social Insurance Administration and subsequent checks of authenticity, is not expected to be completed until 2007, at the earliest. Until this is done, it is possible to provide only rough estimates of total NDC liabilities. Once completed, however, it will be possible for Latvia to implement the balance sheet approach developed and adopted in Sweden. 12 Whether or not this will be done is a new policy issue. In principle, this would be desirable to ensure that system assets are always at least as great as liabilities. In practice, ceteris paribus, as long as there is substantially less than full wage indexation of benefits the system will run a surplus. Hence, given the present design of the system, it is highly unlikely that a brake would be needed in the nearest decades. We return to the issue of future financial stability, given the present legislation, in a separate section below. The FDC Component of the Overall Latvian Reform The FDC scheme is mandatory for persons born after July 1, 1971 that is, persons under the age of 30 on July 2001, when the FDC scheme started operating. Participation is voluntary for persons who at the time of introduction were at that moment in the age group (born between July 2, 1951, and July 1, 1971), with no limiting date on the time at which people can opt into the FDC scheme. According to the current legislation, the contribution rate for participants in the FDC scheme will abide by the following schedule:

11 NDC STRATEGY IN LATVIA: IMPLEMENTATION AND PROSPECTS FOR THE FUTURE 405 from 2 percent in years to 4 percent in 2007, 8 percent in 2008, 9 percent in 2009, and 10 percent from Their NDC contribution rate is reduced by the same schedule as the increase in the FDC rate. Voluntary participation in the FDC scheme was very low in the beginning. However, by the end of 2003, the participation rate of volunteers had increased sharply: to 28 percent from 8 percent at the end of 2002, as private asset managers took over the management of individual accounts and marketed the system. 13 At retirement, participants in the FDC scheme can choose between contracting an annuity on the market and transferring their accumulated financial funds to the NDC reserve fund, 14 allowing them to receive an NDC benefit based on the sum of their NDC and FDC capital. This is a unique feature of the Latvian FDC system. The FDC scheme is expected to be fully mandatory around 2035, when cohorts of voluntary participants will no longer be in the work force. To conclude, as the FDC contribution rate increases, the NDC rate decreases commensurately for participants in the FDC scheme. The implications of the transition for the financial development of the overall system are discussed in a separate section below. The Pension Age and Special Rights Two major reform issues were the low pension age and the widespread use of special rights. How these were dealt with is the topic of this section. The reform concept, passed in 1994, called for an increase in the minimum age for retirement. This established a pension age of 60 for both men and women, which meant increasing the pension age of women from 55 to 60. Later the minimum age was raised to 62 for both men and women. The increase followed a gradual schedule, increasing at the rate of a half year per year. The retirement age became 62 for men already in 2003, but it will not reach 62 for women until Note that in 2002 life expectancy at age 60 was over 15 years for men and almost 22 years for women, suggesting that even a minimum age of 62 is low. Certainly, the issue of the minimum pension age will have to be revisited within the coming decade or so. In the reform legislation of 1995 there are specific paragraphs detailing how special rights for example, those for workers working under hazardous conditions who had a right to a pension at age 40 are to be phased out. This was accomplished by valuing and converting acquired rights in the old regime into NDC capital. The process of phasing out special rights inherited from the old regime went smoothly politically, perhaps because the conversion was perceived as fair, and perhaps because it was seen as fair for everyone to move into the new system on an equal status. On the other hand, as we will discuss in the next section, a new group of special rights was nevertheless legislated after the 1996 reform for persons who were still working once they had passed the full benefit age in the old DB scheme. These turned out to be so blatantly unfair that it is unlikely that this will occur again. Latvians retiring in 2003 had participated 31 years in the labor force, which can be compared with a unisex average of around 35 years for the Western European Union countries in the mid-1990s. 15 More importantly, it is also much lower than it was in Latvia at the time of the reform in 1995, when the average number of covered years was 36 years for the entire stock of pensioners. In 2003 the average for the stock had dropped to 33, however, reflecting the drop in years of newly retired persons during the second half of the 1990s. The number of covered years for a person receiving a newly granted benefit was only 30 in 1996, with an increase to 31 by The drop in years of lifetime labor force participation is probably a temporary effect of the difficult years of the transition, 16 when it became particularly difficult for older workers to move into new jobs, and this provided clear justification for a transition guarantee for older workers (see below). 17

12 406 PENSION REFORM: ISSUES AND PROSPECTS FOR NDC SCHEMES Open Issues Among the open issues in the Latvian pension reform is the issue of what to do with inheritance capital of persons who die before reaching the pension age in both the NDC and FDC schemes. NDC inheritance capital generated by deaths of contributors prior to retirement provides an extra source of finance. Presently in Latvia, it is used to finance the PAYG survivor benefits, which under the new law applies mostly to surviving parents with children under the age of 18 (or up to 24 if they are full-time students). According to the present legislation for the FDC scheme, inheritance capital arising in the FDC scheme is transferred to the social insurance budget, also to finance a mandatory survivor benefit. However, with the present construction of the survivor benefit rules, the benefit is overfinanced from these sources. There are several possible candidates for use of non ear-marked inheritance capital. First, the existing survivor benefit could be improved, which de facto means improving the income status of women (as survivors` pensions are granted for children). This would require a change in the legislation. Second, it could be used to create a better noncontributory childcare credit for women (and men if spouses so choose) to be paid into the NDC and FDC accounts. Third, these gains could be distributed to all surviving participants in each cohort at the minimum pension age in proportion to their share in the total notional capital of the cohort. 18 Fourth, it could be used to cover unfinanced commitments, such as the transition guarantee and, even more generally, the overall cost of transition to the new system both of which are to be financed without increasing the present contribution rate. Implicitly, the latter is the option chosen today, more by default than conscious political design. Conversion of Rights Acquired under the Old System to NDC Capital At the time of the reform, it was estimated that only 65 to 70 percent of earnings were being reported. 19 In fact, for many years in the 1990s, the remaining state-owned enterprises were among the most avid sinners. The transition rules of the new pension law were created with an emphasis on quickly reducing underreporting of earnings and evasion, and thus reducing labor market distortions created by evasion. More generally, the practice of keeping two books was widespread; in fact, no one was penalized directly as long as benefits were not directly related to contributions. Instead, those who were really penalized were those who actually paid full contributions, since by paying for the entire system costs they were subsidizing workers who had the opportunity to evade. A goal of the reform was to introduce the NDC scheme quickly, in the hope that the link between contributions and benefits would discourage underreporting and evasion and in any event tip the fairness scales more in favor of those who paid their contributions. Acquired rights under the old system were to be converted into notional initial capital following the logic of the NDC principle. Accordingly, initial capital was based on the individual s covered wage after 1996, and service years acquired prior to Acquired special rights were valued and converted into initial capital. The general formula in the 1996 legislation for initial capital is: IC = individual covered earnings individual service years The original rule used to determine the individual s covered wage was: the average covered wage for all participants in 1995 for persons retiring in 1996; the individual s own average covered wage in 1996 for retirees in 1997; the individual s own average covered wage in for retirees in 1998; the individual s own average covered wage in for retirees in 1999; and the individual s average covered wage in for persons claiming a benefit in 2000 and later.

13 NDC STRATEGY IN LATVIA: IMPLEMENTATION AND PROSPECTS FOR THE FUTURE 407 The choice to use the individual s own wage after 1995 as the basis for conversion of prereform service years to initial capital was aimed at creating a strong incentive to report income and penalize evasion. 20 This method had its drawbacks, however. The timing and length of the period used to compute this wage proved to be a problem for persons who genuinely had a poor footing in the labor market during the designated conversion period of This meant that persons with similar working histories before 1996 could receive very different pensions based on their employment status after 1995, which could easily be claimed to be unfair particularly to older workers with the vast majority of their years in the old system. On the other hand, in judging this result it is important to keep in mind that there was a dispersion of earnings even before the transition, when most service years were accumulated, and at least to some extent, the distribution of posttransition individual earnings in the initial decade reflected the distribution of pretransition individual earnings. For this reason, the use of a universal average-wage rule would also have been unfair to persons with pre-1996 above average earnings careers. Given the conversion rule chosen, another conversion problem was that some employees did not have a complete contribution record during , either because the employer had defaulted, or because they chose themselves to work informally. In principle, contributions of bankrupt employers should have been covered by the state as a part of the bankruptcy process, even in a transition setting, 22 which is the usual practice in the European Union and other countries outside the transition block. This did not happen in Latvia. Whether persons working under informal circumstances after 1996, who had service records (and rights) under the old regime but consciously defaulted under the new regime, were treated unfairly has no clear-cut answer. It could be argued both ways. However, for some it just was not easy to find regular employment in this phase of the transition, and for this reason the employment-related transition rule for computing the value of capital to be taken forward from the old system could give a poor pension. 23 Because exclusion from the labor force in the first years after the reform in spite of a long prereform contribution record was one of the reasons some initial pensions were too low, there was a need to provide a transition guarantee, and for this reason additional rules were adopted for computing initial capital. These rules were adopted in parliament in 1996 and were applied in , and then again from 2002 to 2010, following a new revision of the law. The transition-guarantee rule applies to persons with at least 30 years of service. According to the guarantee, initial capital is based on the best of either the individual s earnings according to the main law, or an amount based on the average wage for all participants in the years For all others, initial capital is computed as the best of either the computation using the individual s own earnings or an amount based on 40 percent of the average covered wage, which is approximately the minimum wage. Hence, in practice, the minimum guaranteed amount of initial capital based on service years prior to 1996 is that which corresponds to the minimum wage. As time passes, the importance of initial capital will decline, and total notional capital will depend more and more on actual individual earnings and contributions, and less on pre-1996 service years and the transition rule. In sum, the labor market was in such a state of disarray during the transition that the conversion rule in the original law proved to be unfair for persons with long work careers but who were genuinely excluded from the labor market during the latter half of the 1990s. For this reason, a separate guarantee rule was introduced in spite of the existence of a general guaranteed minimum pension that covered rights earned prior to the transition and introduction of the reform. The design of the special guarantee was nevertheless influenced by the goal of minimizing an overly generous transition rule that would

14 408 PENSION REFORM: ISSUES AND PROSPECTS FOR NDC SCHEMES continue to reward underreporting and evasion, an ever-present problem throughout the transition. Individual Benefits the First Decade after Reform How did the reform affect newly granted benefits during its first years? This question can be answered by examining what has happened with the factors that determine an NDC benefit: life expectancy, the average wage, the rate of return applied to accounts, and the number of covered years of new retirees. Increasing real wages and a positive rate of return on accounts mean the yearly increments to NDC capital of younger birth cohorts will exceed those for older birth cohorts. If the average number of covered years remains constant, then development of benefits will depend on the development of unisex life expectancy and the average age at which people choose to retire. We have already seen (table 15.1) that average wage growth was 5.5 percent for all workers, and that the rate of return on NDC accounts was even higher 6.6 percent in the period because of positive growth in the number of persons contributing. This suggests that wage and contribution growth have had a strong positive effect on new benefits. Since 1996, the unisex life expectancy factor at age 60 has increased from 18.25, applied for pension calculation in 1996 to in In 1997, the number of covered years for a newly retired person was 30 years, while in 2003 it was 31. The average age of a new retiree for men in 1997 was 59.8 and for women 55.7 and in 2003 it was respectively 61.1 and 57.7, respectively. Hence, all the factors except life expectancy worked in the direction of increasing the average real value of a newly granted annuity. In fact, the real growth factor was especially strong. Surprisingly, viewed in 2003, the average value of a pension granted in 1996 and each succeeding year thereafter was about the same, with the notable exception of 1999 for men and 1999 and 2000 for women. There are several reasons why the average benefit remained so stable and, together, these disguised the pure effect of the introduction of the NDC scheme. One is that early retirement with an actuarially reduced benefit is possible until Hence, in spite of the increase in the pension age, people could leave the labor force earlier during the initial years, albeit with an actuarially reduced benefit. In practice, this was what many especially many women did. Probably the most important reason why a trend increase in the average real benefit level of new cohorts of pensioners is not observable is that parliament granted extra indexation of benefits in , with especially large changes in 1997 and The overall effect of these extra indexations was to bring benefits granted earlier up to about the level of newly granted benefits. In 1999, the level of an average pension granted was raised by yet another ad hoc measure, this time the introduction of a new special right. Working pensioners whose benefits had been granted under the old law were allowed to recalculate their benefits using the new NDC formula, if they had worked at least three years from This procedure was passed by parliament in response to a powerful interest lobby (those who benefited were older professional workers), in spite of the Ministry of Welfare s objections. In 1995, when the reform was being considered, the ratio of an average newly granted benefit to an average wage was around 44 percent at age 55 for women, and 46 percent at age 60 for men. 26 In 2003, an average newly granted benefit was 67 lats and an average wage was 172 lats in the same year. This gives a ratio of average newly granted benefit to a gross average wage of 39 percent. At the same time pensions (below 100 lats) are not taxed, while wages have an average tax of 25 percent. The ratio of an average newly

15 NDC STRATEGY IN LATVIA: IMPLEMENTATION AND PROSPECTS FOR THE FUTURE 409 granted benefit to an average after-tax wage is thus about 49 percent. Overall, the effect of the life expectancy factor and the decline in the number of average covered years during the initial transition decade have tended to reduce the level of newly granted benefits. With a more stable economic environment and increase in the minimum pension age, the number of covered years will increase. Summing up, a low number of covered years on average together with a pronounced tendency for people to avail an option to retire before the minimum pension age until July 2005 contributed to holding down the average size of newly granted pensions in the initial years. In addition, extra indexations raised the benefit levels of the pre- and early postreform pensioners to a level close to newly granted benefits in the following years. NDC accounts have grown with the rapid increase in the per capita wage, but the average replacement rate is nevertheless slightly lower than immediately prior to the reform, largely because of the low number of covered years. It is reasonable to believe that the number of years of coverage will once again return to a higher level the average number of covered years was around 35 for the whole stock of pensioners at the time the reform was being formulated in Both longer careers with covered earnings and adaptation to the new higher retirement age will contribute to higher pensions in the future. The Long-Term Financial Picture The primary goal of the 1996 Latvian pension reform was to create an adequate benefit for career workers through a system that links benefits to contributions within the framework of a system that is financially stable in the long run. This was the reasoning behind the choice of a combination of NDC and an FDC. 27 The introduction of NDC and FDC also provides a framework for minimizing labor-market distortions and creating savings, while in the long run this new framework can contribute to higher economic growth and prosperity for all. In this section we analyze the long-term financial picture for this choice. We begin by discussing demographic and economic scenarios for the coming half century, and then show how the financial development of the new pension system is affected by more extreme assumptions about economic growth, fertility, morality, and net immigration. The Demographic and Economic Setting The future development of fertility, mortality, and net migration are all extremely uncertain in any country. Latvia is no exception. Latvia experienced a baby boom in the 1980s, but since the beginning of the transition the fertility rate has been persistently low. In 2003, the period fertility rate was only 1.2 children per woman, far below the 2.1 required to maintain a stable population. In addition, there is considerable uncertainty regarding the effect of EU membership on net migration. In the medium term, there is no doubt that many Latvians, particularly younger and better-educated ones, will seek economic opportunities outside Latvia. Both the strength and duration of this effect of the transition is difficult to judge, however. Furthermore, there is considerable room for mortality to drop dramatically in the coming decades if diets, smoking, and living habits in general become healthier, and as modern medical technology becomes more accessible and affordable. 28 The question addressed here is how financially resilient the new Latvian system is in the face of extreme demographic and, hence, demographically generated economic pressure. In order to examine the resilience of the reform to various constellations of parameters, a number of scenarios some with very extreme economic and demographic

16 410 PENSION REFORM: ISSUES AND PROSPECTS FOR NDC SCHEMES assumptions are examined. The baseline scenario assumes an increase in fertility from the 2003 level of 1.2 to 1.5 in 2015, and then to 1.7 for the remainder of the period examined. An even more optimistic scenario, with period fertility reaching 1.8 in 2015 and then increasing and maintaining a level of 2.0 assuming optimistically that economic prosperity will increase the propensity to have children is also examined. The baseline scenario assumes net migration from Latvia through Thereafter, net migration is zero. Two other scenarios for net migration include positive and negative net immigration flows after Two mortality scenarios are examined. In the baseline scenario, survival of men from age 60 increases from 15.1 in 2002 to 18.1 years in 2050, and for women from 21.8 in 2002 to 23.5 years in A set of more dramatic changes is also examined, with an increase in survival for men age 60 to 23.3 in year 2050 and for women age 60 to This more dramatic change assumes more than an 8-year increase in longevity for men and more than a 6-year increase for women. This would put Latvians on a level closer to their Scandinavian neighbors. As figures 15.2 through 15.4 show, in the baseline demography the total Latvian population declines from about 2.3 million persons in 2003 to 1.5 million persons in year 2075, which assumes that the present net migration out of the country stops by In the scenario with the continued net migration of younger persons out of the country, the population falls to less than 1.3 million persons in The yearly rate of decline in the population is 0.57 per year in the baseline, where net emigration stops in 2010, and 0.76 per year in the pessimistic scenario, where it continues. The pessimistic scenario embodies, thus, high migration and low fertility and reduces the population dramatically. Figure The Latvian Population 4,000 3,500 3,000 2,500 2,000 1,500 1, thousands of persons baseline baseline with low mortality baseline with high fertility baseline with high fertility and high immigration baseline with high emigration Source: Authors calculations using the Latvian Ministry of Welfare s Technical Model.